From AI infrastructure to intelligent experience

As enterprises require greater computing capacity for model training, inference and AI applications, AI infrastructure has become a concern. TechTrade Asia discussed the growing role of AI infrastructure with Bull Hsu, Director, Ubitus (BH):

Q. Why is AI infrastructure becoming a strategic technology layer?

BH: AI infrastructure is becoming increasingly important as organisations move from experimenting with AI to deploying it in real-world applications. Over the past year, we have seen a shift from asking whether AI can be used to recognising what it can actually do, from supporting software development to operating as an agent and interacting with systems. 

As adoption grows, enterprises need greater GPU capacity and scalable computing environments, but they also need to consider latency, privacy and where their data is stored. This is driving greater interest in local and sovereign AI infrastructure, particularly among organisations that want to keep sensitive data within their own country or controlled environment.

Q: How will this affect the next phase of AI adoption in Asia? 

A: The next phase of AI adoption in Asia will depend increasingly on access to reliable computing capacity. Organisations are no longer simply testing whether AI works; they are looking at how to deploy it using their own data and how to tailor AI to the specific needs of their businesses and industries. 

This is particularly important in sectors such as healthcare and financial services, where sensitive information needs to remain within controlled environments.

Q: How can AI infrastructure support new customer experiences? 

BH: The value of AI infrastructure ultimately comes from what it enables people to do. By combining GPU computing, AI models and application platforms, organisations can create virtual characters, digital assistants and interactive services that respond in real time and deliver more personalised experiences.

For example, Ubitus and its partners in Taiwan are working with a hospital on virtual AI characters that can interact with children. These virtual characters can make interactions more approachable while providing information about health and other areas.

In entertainment, we are also developing AI virtual YouTubers that can manage social platforms, interact with fans and play games. These applications demonstrate how AI is moving beyond simple chatbots towards agents that can perform multiple tasks. 

Responsiveness is also critical. People expect natural interactions, so our experience developing AI virtual characters has focused on reducing latency and enabling faster, more natural conversations.

Q: Where could AI-powered interfaces and immersive experiences develop next across telecoms, media, advertising, tourism, entertainment and enterprise services? 

BH: AI interfaces are moving beyond traditional text and voice towards more natural and immersive ways of interacting with technology. AI virtual humans, digital assistants and AI agents could become new interfaces for customer service, tourism, education, entertainment and enterprise services. These applications could include virtual ambassadors and guides, as well as AI agents that interact with users, manage digital platforms and carry out tasks on their behalf.

The foundation comes first: sufficient computing power, reliable infrastructure, local data environments and low latency. With that foundation in place, organisations can build more sophisticated AI services and interactive experiences for customers across Asia.  

Editor's note: Ubitus is responding to industry shifts by providing AI infrastructure, platforms, models and AI solutions
that help customers build the computing foundation needed to deploy AI at scale. In particular, Ubitus is expanding its GPU computing environment across Taiwan, Japan and Malaysia.